/external/XNNPACK/src/qs8-vadd/gen/ |
D | minmax-neon-ld64-x32.c | 52 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x32() 53 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x32() 54 int32x4_t vacc89AB = vmulq_s32(vmovl_s16(vget_low_s16(vxa89ABCDEF)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x32() 55 int32x4_t vaccCDEF = vmulq_s32(vmovl_s16(vget_high_s16(vxa89ABCDEF)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x32() 56 int32x4_t vaccGHIJ = vmulq_s32(vmovl_s16(vget_low_s16(vxaGHIJKLMN)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x32() 57 int32x4_t vaccKLMN = vmulq_s32(vmovl_s16(vget_high_s16(vxaGHIJKLMN)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x32() 58 int32x4_t vaccOPQR = vmulq_s32(vmovl_s16(vget_low_s16(vxaOPQRSTUV)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x32() 59 int32x4_t vaccSTUV = vmulq_s32(vmovl_s16(vget_high_s16(vxaOPQRSTUV)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x32() 104 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x32() 105 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x32()
|
D | minmax-neon-ld128-x32.c | 64 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x32() 65 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x32() 66 int32x4_t vacc89AB = vmulq_s32(vmovl_s16(vget_low_s16(vxa89ABCDEF)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x32() 67 int32x4_t vaccCDEF = vmulq_s32(vmovl_s16(vget_high_s16(vxa89ABCDEF)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x32() 68 int32x4_t vaccGHIJ = vmulq_s32(vmovl_s16(vget_low_s16(vxaGHIJKLMN)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x32() 69 int32x4_t vaccKLMN = vmulq_s32(vmovl_s16(vget_high_s16(vxaGHIJKLMN)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x32() 70 int32x4_t vaccOPQR = vmulq_s32(vmovl_s16(vget_low_s16(vxaOPQRSTUV)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x32() 71 int32x4_t vaccSTUV = vmulq_s32(vmovl_s16(vget_high_s16(vxaOPQRSTUV)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x32() 121 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x32() 122 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x32()
|
D | minmax-neon-ld64-x24.c | 48 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x24() 49 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x24() 50 int32x4_t vacc89AB = vmulq_s32(vmovl_s16(vget_low_s16(vxa89ABCDEF)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x24() 51 int32x4_t vaccCDEF = vmulq_s32(vmovl_s16(vget_high_s16(vxa89ABCDEF)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x24() 52 int32x4_t vaccGHIJ = vmulq_s32(vmovl_s16(vget_low_s16(vxaGHIJKLMN)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x24() 53 int32x4_t vaccKLMN = vmulq_s32(vmovl_s16(vget_high_s16(vxaGHIJKLMN)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x24() 93 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x24() 94 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x24()
|
D | minmax-neon-ld64-x16.c | 44 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x16() 45 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x16() 46 int32x4_t vacc89AB = vmulq_s32(vmovl_s16(vget_low_s16(vxa89ABCDEF)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x16() 47 int32x4_t vaccCDEF = vmulq_s32(vmovl_s16(vget_high_s16(vxa89ABCDEF)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x16() 78 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x16() 79 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x16()
|
D | minmax-neon-ld128-x16.c | 54 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x16() 55 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x16() 56 int32x4_t vacc89AB = vmulq_s32(vmovl_s16(vget_low_s16(vxa89ABCDEF)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x16() 57 int32x4_t vaccCDEF = vmulq_s32(vmovl_s16(vget_high_s16(vxa89ABCDEF)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x16() 93 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x16() 94 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld128_x16()
|
D | minmax-neon-ld64-x8.c | 40 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x8() 41 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x8() 67 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x8() 68 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qs8_vadd_minmax_ukernel__neon_ld64_x8()
|
/external/XNNPACK/src/qu8-vadd/gen/ |
D | minmax-neon-ld64-x32.c | 52 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x32() 53 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x32() 54 int32x4_t vacc89AB = vmulq_s32(vmovl_s16(vget_low_s16(vxa89ABCDEF)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x32() 55 int32x4_t vaccCDEF = vmulq_s32(vmovl_s16(vget_high_s16(vxa89ABCDEF)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x32() 56 int32x4_t vaccGHIJ = vmulq_s32(vmovl_s16(vget_low_s16(vxaGHIJKLMN)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x32() 57 int32x4_t vaccKLMN = vmulq_s32(vmovl_s16(vget_high_s16(vxaGHIJKLMN)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x32() 58 int32x4_t vaccOPQR = vmulq_s32(vmovl_s16(vget_low_s16(vxaOPQRSTUV)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x32() 59 int32x4_t vaccSTUV = vmulq_s32(vmovl_s16(vget_high_s16(vxaOPQRSTUV)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x32() 104 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x32() 105 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x32()
|
D | minmax-neon-ld64-x16.c | 44 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x16() 45 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x16() 46 int32x4_t vacc89AB = vmulq_s32(vmovl_s16(vget_low_s16(vxa89ABCDEF)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x16() 47 int32x4_t vaccCDEF = vmulq_s32(vmovl_s16(vget_high_s16(vxa89ABCDEF)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x16() 78 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x16() 79 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x16()
|
D | minmax-neon-ld128-x16.c | 54 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld128_x16() 55 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld128_x16() 56 int32x4_t vacc89AB = vmulq_s32(vmovl_s16(vget_low_s16(vxa89ABCDEF)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld128_x16() 57 int32x4_t vaccCDEF = vmulq_s32(vmovl_s16(vget_high_s16(vxa89ABCDEF)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld128_x16() 93 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld128_x16() 94 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld128_x16()
|
D | minmax-neon-ld64-x8.c | 40 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x8() 41 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x8() 67 int32x4_t vacc0123 = vmulq_s32(vmovl_s16(vget_low_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x8() 68 int32x4_t vacc4567 = vmulq_s32(vmovl_s16(vget_high_s16(vxa01234567)), va_multiplier); in xnn_qu8_vadd_minmax_ukernel__neon_ld64_x8()
|
/external/libhevc/common/arm/ |
D | ihevc_resi_trans_neon_32x32.c | 1203 a[0] = vmulq_s32(g_ai4_ihevc_trans_32_0_8, eeee); in ihevc_resi_trans_32x32_neon() 1215 a[4] = vmulq_s32(g_ai4_ihevc_trans_32_4_04, eeo); in ihevc_resi_trans_32x32_neon() 1216 a[12] = vmulq_s32(g_ai4_ihevc_trans_32_12_04, eeo); in ihevc_resi_trans_32x32_neon() 1217 a[20] = vmulq_s32(g_ai4_ihevc_trans_32_20_04, eeo); in ihevc_resi_trans_32x32_neon() 1218 a[28] = vmulq_s32(g_ai4_ihevc_trans_32_28_04, eeo); in ihevc_resi_trans_32x32_neon() 1245 a[2] = vmulq_s32(g_ai4_ihevc_trans_32_2_03, eo0); //q8 in ihevc_resi_trans_32x32_neon() 1246 a[6] = vmulq_s32(g_ai4_ihevc_trans_32_6_03, eo0); //q2 in ihevc_resi_trans_32x32_neon() 1247 a[10] = vmulq_s32(g_ai4_ihevc_trans_32_10_03, eo0); //q2 in ihevc_resi_trans_32x32_neon() 1248 a[14] = vmulq_s32(g_ai4_ihevc_trans_32_14_03, eo0); //q2 in ihevc_resi_trans_32x32_neon() 1275 a[18] = vmulq_s32(g_ai4_ihevc_trans_32_18_03, eo0); //q2 in ihevc_resi_trans_32x32_neon() [all …]
|
D | ihevc_quant_iquant_ssd_neon_intr.c | 280 r0 = vmulq_s32(r0, r0); in ihevc_quant_iquant_ssd_flat_scale_mat_neon() 281 r1 = vmulq_s32(r1, r1); in ihevc_quant_iquant_ssd_flat_scale_mat_neon() 282 r2 = vmulq_s32(r2, r2); in ihevc_quant_iquant_ssd_flat_scale_mat_neon() 283 r3 = vmulq_s32(r3, r3); in ihevc_quant_iquant_ssd_flat_scale_mat_neon() 640 r0 = vmulq_s32(r0, r0); in ihevc_q_iq_ssd_flat_scale_mat_var_rnd_fact_neon() 641 r1 = vmulq_s32(r1, r1); in ihevc_q_iq_ssd_flat_scale_mat_var_rnd_fact_neon() 642 r2 = vmulq_s32(r2, r2); in ihevc_q_iq_ssd_flat_scale_mat_var_rnd_fact_neon() 643 r3 = vmulq_s32(r3, r3); in ihevc_q_iq_ssd_flat_scale_mat_var_rnd_fact_neon()
|
D | ihevc_resi_trans_neon.c | 281 int32x4_t c3_4x32b = vmulq_s32(vmovl_s16(src2_4x16b), coeff2_4x32b); /* 74*r2 */ in ihevc_resi_trans_4x4_ttype1_neon() 286 src2_4x32b = vmulq_s32(c2_4x32b, coeff0_4x32b); /* 29*c2 - c3 */ in ihevc_resi_trans_4x4_ttype1_neon() 292 c1_4x32b = vmulq_s32(c1_4x32b, coeff0_4x32b); /* 55*c2 - 29*c1 + c3 */ in ihevc_resi_trans_4x4_ttype1_neon() 293 src1_4x32b = vmulq_s32(src1_4x32b, coeff2_4x32b); /*74*(r0+r1-r3)*/ in ihevc_resi_trans_4x4_ttype1_neon() 319 c3_4x32b = vmulq_s32(src2_4x32b, coeff2_4x32b); /* 74*r2 */ in ihevc_resi_trans_4x4_ttype1_neon() 324 src2_4x32b = vmulq_s32(c2_4x32b, coeff0_4x32b); /* 29*c2 - c3 */ in ihevc_resi_trans_4x4_ttype1_neon() 330 c1_4x32b = vmulq_s32(c1_4x32b, coeff0_4x32b); /* 55*c2 - 29*c1 + c3 */ in ihevc_resi_trans_4x4_ttype1_neon() 331 src1_4x32b = vmulq_s32(src1_4x32b, coeff2_4x32b); /*74*(r0+r1-r3)*/ in ihevc_resi_trans_4x4_ttype1_neon()
|
/external/libvpx/vpx_dsp/arm/ |
D | highbd_quantize_neon.c | 88 dqcoeff_0 = vmulq_s32(qcoeff_0, dequant); in highbd_quantize_b_neon() 89 dqcoeff_1 = vmulq_s32(qcoeff_1, vdupq_lane_s32(vget_low_s32(dequant), 1)); in highbd_quantize_b_neon() 218 dqcoeff_0 = vmulq_s32(qcoeff_0, dequant); in highbd_quantize_b_32x32_neon() 219 dqcoeff_1 = vmulq_s32(qcoeff_1, vdupq_lane_s32(vget_low_s32(dequant), 1)); in highbd_quantize_b_32x32_neon()
|
/external/libhevc/encoder/arm/ |
D | ihevce_common_utils_neon.c | 139 reg0[0] = vmulq_s32(reg0[0], a6); in ihevce_wt_avg_2d_16x1_neon() 140 reg0[1] = vmulq_s32(reg0[1], a6); in ihevce_wt_avg_2d_16x1_neon() 141 reg0[2] = vmulq_s32(reg0[2], a6); in ihevce_wt_avg_2d_16x1_neon() 142 reg0[3] = vmulq_s32(reg0[3], a6); in ihevce_wt_avg_2d_16x1_neon() 144 reg1[0] = vmulq_s32(reg1[0], a7); in ihevce_wt_avg_2d_16x1_neon() 145 reg1[1] = vmulq_s32(reg1[1], a7); in ihevce_wt_avg_2d_16x1_neon() 146 reg1[2] = vmulq_s32(reg1[2], a7); in ihevce_wt_avg_2d_16x1_neon() 147 reg1[3] = vmulq_s32(reg1[3], a7); in ihevce_wt_avg_2d_16x1_neon() 198 a8 = vmulq_s32(a8, a4); in ihevce_wt_avg_2d_8x1_neon() 199 a9 = vmulq_s32(a9, a4); in ihevce_wt_avg_2d_8x1_neon() [all …]
|
/external/XNNPACK/src/qs8-vadd/ |
D | neon.c.in | 111 … int32x4_t vacc${ABC[N:N+4]} = vmulq_s32(vmovl_s16(vget_low_s16(vxa${ABC[N:N+8]})), va_multiplier); 112 …int32x4_t vacc${ABC[N+4:N+8]} = vmulq_s32(vmovl_s16(vget_high_s16(vxa${ABC[N:N+8]})), va_multiplie… 186 int32x4_t vacc${ABC[0:4]} = vmulq_s32(vmovl_s16(vget_low_s16(vxa${ABC[0:8]})), va_multiplier); 187 … int32x4_t vacc${ABC[4:8]} = vmulq_s32(vmovl_s16(vget_high_s16(vxa${ABC[0:8]})), va_multiplier);
|
/external/ComputeLibrary/src/core/NEON/ |
D | NEAsymm.h | 263 vmulq_s32(in_s32.val[0], vshlq_s32(one_s32, vnegq_s32(result_shift.val[0]))), in finalize_quantization_symm() 264 vmulq_s32(in_s32.val[1], vshlq_s32(one_s32, vnegq_s32(result_shift.val[1]))), in finalize_quantization_symm() 265 vmulq_s32(in_s32.val[2], vshlq_s32(one_s32, vnegq_s32(result_shift.val[2]))), in finalize_quantization_symm() 266 vmulq_s32(in_s32.val[3], vshlq_s32(one_s32, vnegq_s32(result_shift.val[3]))), in finalize_quantization_symm()
|
/external/libaom/av1/common/arm/ |
D | warp_plane_neon.c | 646 res_lo = vmulq_s32(res_lo, bwd); in av1_warp_affine_neon() 647 tmp32_lo = vmulq_s32(tmp32_lo, fwd); in av1_warp_affine_neon() 677 res_hi = vmulq_s32(res_hi, bwd); in av1_warp_affine_neon() 678 tmp32_hi = vmulq_s32(tmp32_hi, fwd); in av1_warp_affine_neon()
|
D | selfguided_neon.c | 1073 a_res0 = vmulq_s32(vmovl_s16(vget_low_s16(s0)), a_res0); in final_filter_fast_internal() 1074 a_res1 = vmulq_s32(vmovl_s16(vget_high_s16(s0)), a_res1); in final_filter_fast_internal() 1099 a_res0 = vmulq_s32(vmovl_s16(vget_low_s16(s0)), a_res0); in final_filter_fast_internal() 1100 a_res1 = vmulq_s32(vmovl_s16(vget_high_s16(s0)), a_res1); in final_filter_fast_internal() 1151 a_res0 = vmulq_s32(vmovl_s16(vget_low_s16(s0)), a_res0); in final_filter_internal() 1152 a_res1 = vmulq_s32(vmovl_s16(vget_high_s16(s0)), a_res1); in final_filter_internal()
|
/external/libaom/av1/encoder/arm/neon/ |
D | av1_highbd_quantize_neon.c | 45 vshlq_s32(vmulq_s32(v_abs_qcoeff, v_dequant_s32), vnegq_s32(v_log_scale)); in quantize_4()
|
/external/ComputeLibrary/src/cpu/kernels/ |
D | CpuGemmLowpOffsetContributionOutputStageKernel.cpp | 119 vmulq_s32(a.val[0], vld1q_s32(multilpier)), in mul_s32() 120 vmulq_s32(a.val[1], vld1q_s32(multilpier + 4)), in mul_s32() 121 vmulq_s32(a.val[2], vld1q_s32(multilpier + 8)), in mul_s32() 122 vmulq_s32(a.val[3], vld1q_s32(multilpier + 12)) in mul_s32()
|
/external/libvpx/vp9/encoder/arm/neon/ |
D | vp9_quantize_neon.c | 248 const int32x4_t v_abs_dqcoeff = vmulq_s32(v_abs_qcoeff, v_dequant_s32); in highbd_quantize_fp_4() 339 vshrq_n_s32(vmulq_s32(v_abs_qcoeff, v_dequant_s32), 1); in highbd_quantize_fp_32x32_4()
|
/external/libaom/aom_dsp/arm/ |
D | highbd_quantize_neon.c | 46 const int32x4_t v_tmpw32 = vmulq_s32(v_tmp, vdupq_n_s32((1 << AOM_QM_BITS))); in quantize_4() 59 vshlq_s32(vmulq_s32(v_abs_qcoeff, v_dequant_s32), vnegq_s32(v_log_scale)); in quantize_4()
|
/external/libgav1/src/dsp/arm/ |
D | motion_vector_search_neon.cc | 39 const int32x4_t m = vmulq_s32(m0, numerator); in MvProjection()
|
/external/tensorflow/tensorflow/lite/kernels/internal/optimized/ |
D | neon_tensor_utils.cc | 184 vmulq_s32(input_val.val[0], left_shifted_one_dup), in MultiplyByQuantizedMultiplier2Rows() 189 vmulq_s32(input_val.val[1], left_shifted_one_dup), in MultiplyByQuantizedMultiplier2Rows() 1537 AccumulateNeonLane(vmulq_s32(val_s32_0, val_s32_0))); in NeonApplyLayerNorm() 1539 AccumulateNeonLane(vmulq_s32(val_s32_1, val_s32_1))); in NeonApplyLayerNorm() 1758 int32x4_t x_0 = vmulq_s32(a_s32_0, b_s32_0); in NeonCwiseMul() 1759 int32x4_t x_1 = vmulq_s32(a_s32_1, b_s32_1); in NeonCwiseMul() 1803 temp_val.val[0] = vmulq_s32(a_s32_0, b_s32_0); in NeonCwiseMul() 1804 temp_val.val[1] = vmulq_s32(a_s32_1, b_s32_1); in NeonCwiseMul()
|